RDA for Automatic Airport Recognition on FLIR Image

This paper studies Regularized Discriminant Analysis (RDA) in the context of automatic airport recognition system for Forward-Looking infrared images (FLIR). When the within class covariance of training sample are sometimes singular, Linear and Quadratic discriminant analysis (LDA & QDA) does not necessarily give the best performance. Alternatives to the usual plug-in (maximum likelihood) estimates for the covariance matrices are proposed, which is called two-parameter RDA in this paper. Here, we check two-parameter RDA availability and compare its performance in our recognition system to other classifiers, such as KNN, LDA, QDA etc. The experimental results demonstrate the efficacy of the two parameters RDA classifier for automatic airport recognition in FUR images. On the basis of RDA classifier, our proposed recognition system frame was concluded to be a highly prospective candidate for real time ATR system on airport and can also be used on other ATR system, such as building, power plant etc.